Small Device C Compiler suite

Hi Philipp,

> Dear sdcc developers,
> 
> the Z180 port in the z180 branch is doing well, and I'd like to merge it
> soon.
> 
> Current state:
> - - Mostly identical to the Z80 port (thus inherits features like full
> bool support and the optimal register allocator)
> - - Can use the Z180 mlt instructions for multiplication (this often
> results in quite different code than for the Z80 for functions that use
> multiplication, since the register allocator rearranges local variables
> to put the multiplication operands inthe right places)
> - - Passes regression tests.

That sounds good. Did you have to modify the simulator 
as well or was the Z180 already supported? Anyway, from 
your description I have no objections to the merge.

Maarten

On 09/24/2011 10:24 AM, Maarten Brock wrote:
> Hi Philipp,
>
>> Dear sdcc developers,
>>
>> the Z180 port in the z180 branch is doing well, and I'd like to merge it
>> soon.
>>
>> Current state:
>> - - Mostly identical to the Z80 port (thus inherits features like full
>> bool support and the optimal register allocator)
>> - - Can use the Z180 mlt instructions for multiplication (this often
>> results in quite different code than for the Z80 for functions that use
>> multiplication, since the register allocator rearranges local variables
>> to put the multiplication operands inthe right places)
>> - - Passes regression tests.
> That sounds good. Did you have to modify the simulator
> as well or was the Z180 already supported? Anyway, from
> your description I have no objections to the merge.

Philipp, what about bug #3413437? Is the "Invalid instruction****" 
caused by bug in the simulator? Or is the Z180 specific mlt instruction, 
not recognized by Z80 simulator?
I also don't have objections to merge once this is clarified.

Borut

Am 24.09.2011 12:20, schrieb Borut Ražem:
> Philipp, what about bug #3413437? Is the "Invalid instruction****"
> caused by bug in the simulator?

The problem that resulted in the invalid instruction has been fixed in
revision #6873.

Philipp

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Am 24.09.2011 10:24, schrieb Maarten Brock:
> Hi Philipp,
> 
>> Dear sdcc developers,
>>
>> the Z180 port in the z180 branch is doing well, and I'd like to merge it
>> soon.
>>
>> Current state:
>> - - Mostly identical to the Z80 port (thus inherits features like full
>> bool support and the optimal register allocator)
>> - - Can use the Z180 mlt instructions for multiplication (this often
>> results in quite different code than for the Z80 for functions that use
>> multiplication, since the register allocator rearranges local variables
>> to put the multiplication operands inthe right places)
>> - - Passes regression tests.
> 
> That sounds good. Did you have to modify the simulator 
> as well or was the Z180 already supported? Anyway, from 
> your description I have no objections to the merge.

The simulator does yet fully support the Z180: I have added support for
the mlt bc, mlt de and mlt hl instructions, since sdcc generates them
when using -mz180. The simulator can be run in Z180 mode using -tz180
(Iused that way of invoking z180 mode, since it seems to be consistent
with what the simulator for mcs51-variants does). I have not yet added
support for the other Z180-specific instructions to the simulator.

Philipp
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAk59yDcACgkQbtUV+xsoLppjZACg7rLysMdhEuFobCCuT7486Mud
C7QAoJCFcTeClclvDZEB8zdILvtdTfxf
=dCia
-----END PGP SIGNATURE-----

Philipp,

the file z180.h, currently located at device/include/z80/z180.h should 
be probably moved to device/include/z180/z180.h?

Borut


On 09/24/2011 02:08 PM, Philipp Klaus Krause wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Am 24.09.2011 10:24, schrieb Maarten Brock:
>> Hi Philipp,
>>
>>> Dear sdcc developers,
>>>
>>> the Z180 port in the z180 branch is doing well, and I'd like to merge it
>>> soon.
>>>
>>> Current state:
>>> - - Mostly identical to the Z80 port (thus inherits features like full
>>> bool support and the optimal register allocator)
>>> - - Can use the Z180 mlt instructions for multiplication (this often
>>> results in quite different code than for the Z80 for functions that use
>>> multiplication, since the register allocator rearranges local variables
>>> to put the multiplication operands inthe right places)
>>> - - Passes regression tests.
>> That sounds good. Did you have to modify the simulator
>> as well or was the Z180 already supported? Anyway, from
>> your description I have no objections to the merge.
> The simulator does yet fully support the Z180: I have added support for
> the mlt bc, mlt de and mlt hl instructions, since sdcc generates them
> when using -mz180. The simulator can be run in Z180 mode using -tz180
> (Iused that way of invoking z180 mode, since it seems to be consistent
> with what the simulator for mcs51-variants does). I have not yet added
> support for the other Z180-specific instructions to the simulator.
>
> Philipp
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.11 (GNU/Linux)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
>
> iEYEARECAAYFAk59yDcACgkQbtUV+xsoLppjZACg7rLysMdhEuFobCCuT7486Mud
> C7QAoJCFcTeClclvDZEB8zdILvtdTfxf
> =dCia
> -----END PGP SIGNATURE-----
>
> ------------------------------------------------------------------------------
> All of the data generated in your IT infrastructure is seriously valuable.
> Why? It contains a definitive record of application performance, security
> threats, fraudulent activity, and more. Splunk takes this data and makes
> sense of it. IT sense. And common sense.
> http://p.sf.net/sfu/splunk-d2dcopy2
> _______________________________________________
> sdcc-devel mailing list
> sdcc-devel@...
> https://lists.sourceforge.net/lists/listinfo/sdcc-devel
>

Philipp,

while investigating the win32 & win64 regtest failures, I foung the 
following lines in SDCCmain.c, function setLibPath():

       /* Use common library for z80 and z180. */
       targetname = strcmp(port->target, "z180") ? port->target : "z80";

Is this rellay needed? I think that the z180.lib library is built, so 
why not using it?

I think that this makes a problem only on win builds. On *nix builds the 
z180 library is found due to the rule:
    * 8. - DATADIR/NON_FREE_LIB_DIR_SUFFIX/<model> (only on *nix)

Borut

Am 09.10.2011 15:02, schrieb Borut Ražem:
> Philipp,
> 
> while investigating the win32 & win64 regtest failures, I foung the 
> following lines in SDCCmain.c, function setLibPath():
> 
>        /* Use common library for z80 and z180. */
>        targetname = strcmp(port->target, "z180") ? port->target : "z80";
> 
> Is this rellay needed? I think that the z180.lib library is built, so 
> why not using it?
> 

It was needed when the z180 port still used the z80 library, but then I
changed the z180 library handling and forgot about this since it didn't
affect regression tests on GNU/Linux. I just fixed it.

Philipp

Am 09.10.2011 14:30, schrieb Borut Ražem:
> Philipp,
> 
> the file z180.h, currently located at device/include/z80/z180.h should 
> be probably moved to device/include/z180/z180.h?

I agree, and will do it and remove the device/include/z80 directory.

Philipp